Category Mapping!

Question

If you have not seen it yet, Aleš recently created a mapping function through WikiTree+. You can read more about the maps here.

Where this really excites me, is through the possibilities to expand research on location and topical categories through the Category Mapping!

Did you know that you can now pull up a specific cemetery category and view all of the profiles through their birth, migration, and death events - directly from a Category that has a CategoryInfoBox?

WikiTree+ Map for Category:Sterling-White Cemetery

Ready to see it in action yourself? Navigate to a category and look for the CategoryInfoBox, then select WikiTree+ Maps link by the Map of Profiles heading! It is that simple!

Ready to learn more? You can find tons of Location, Cemetery, Migration, and One Name Study categories with a CategoryInfoBox ready to map our ancestors!

Don't see a CategoryInfoBox on a category? Have no fear! At the top of the page there are links for WikiTree+ Search / Navigate. Click on the Navigate link, and then click the map icon for the category:

Answer 1

Wow! Definitely good news. Does Wikitree have a super-hero badge to give to Aleš?

My personal interest is the possibility of resolving a recent debate about bridging the gap between:  1) Old location names from family records  and 2) Current location.

Answer 2

I have noticed a bug in this function that is putting the place of birth and death for people who were born or died in New Zealand into a different continent. It is setting the birth or death marker in New Brunswick.

A couple of example profiles are Lawrence-10858, Butler-14595 and Hobbs-2152 .

It appears to only be picking up the last word (Zealand) and assigning the markers to Zealand, New Brunswick, Canada.

I checked some other countries quickly and have found another profile that is miss-assigned which died in West Lothian, Scotland but the marker puts her death place in South America. That profile is Breatcliff-2 .

Comments

You are missing commas in the location. 

Dunedin Otago New Zealand should be entered as Dunedin, Otago, New Zealand  

In case there are no commas in the location, I try to simulate them by replacing all spaces with comma, but the results are not good if there should be a space in the name.

---

In case of Breatcliff, you have a bullet in front of the location. Also country is missing in both locations.

There are multiple places with the same name in a lot of cases in the world. So you must specify a country, to get the correct one.

---

The problem is that the country is in the location field yet the mapping function is adding incorrect information of New Brunswick, Canada. If it followed the same logic of Google maps or OpenStreetMap for examples it should by using the search term Zealand show a location in Denmark instead of a location in New Brunswick, Canada. 

It is also inconsistent as I have found a profile that has the death location as New York and going by the logic of only using the last word it should show the death location as York, England yet it appears in New York, United States of America. 

Programming in a list of Countries at the least to check the location of where the countries have two or more words in them shouldn't be a hard thing to program into the verification process. 

---

I have built many exceptions to the algorithm to automatically improve the matching.

I wouldn't go into all the details, since it is very complicated and above the needs of average WikiTree user.

But here are some details, since you are interested.

First I check if the whole location is a known country (that covers "New York" and "New Zealand"). Then if there is no comma in the location field, I replace spaces with commas. This works perfectly if there are no spaces in any name and mostly there aren't. I can see if I can built a few special cases for NZ and a few others.

But location should have commas in correct places.

There are 25000 unique locations for NZ. 2587 of them are not matched to the NZ. I will see if I can improve that.

---

I recalculated cca 18000 locations based on additional space checks. Can you have a look if those locations are better. cca 1000 of them are NZ related.